Issue with SNP Effect Calculation in GAPIT 3.4

[PetrosRoth]

Dear Dr. Jiabo Wang,

I have been using GAPIT 3.4 to perform GWAS in a wheat panel and have encountered an issue regarding the SNP effect calculation.

After reading the user manual, I set the major.allele.zero argument to TRUE. However, I have found several cases where, although the minor allele shows a beneficial effect based on the distribution plots, the calculated effect size is negative. According to the manual, setting major.allele.zero = TRUE should mean that the sign of the allelic effect estimate is with respect to the minor allele. In other words, a positive allelic effect estimate should indicate that the minor allele is favorable. However, in the case I am attaching below, this is not the case.

For context, I am importing my genotypic data in Hapmap format. Everything else appears to work correctly.

Could you please provide guidance on this issue?

Thank you in advance for your kind help.

Best regards,
Petros

[jiabowang]

Hi Petros,
Thanks for your report.
Recently, we also have observed the mistake in converting HapMap genotype to numeric genotype. In some case, it does not work. So we fixed this BUG.
Could you tell me whether you used the newest GAPIT function?
Please source("https://github.com/jiabowang/GAPIT/blob/master/gapit_functions.txt")
Now this link is same as ZZLAB GAPIT link.
After converting, you can use below code to validate.

myGAPIT<-GAPIT(G=myG,PCA.total=0,kinship.algorithm="",Major.allele.zero = T,file.output=F)# or Major.allele.zero = F
myGD=myGAPIT$GD
i=9 # this number can be random set
table(myGD[,i])
table(as.character(myG[i,-c(1:11)]))

In our case, the genotype were accurately converted.

[PetrosRoth]

Thank you so much for your reply.

So far, we have not converted the HapMap genotype to numeric format first. Instead, we directly called the GAPIT function using the following code:

gwas_results <- GAPIT(
Y = phenotype_data,
G = gen_data, # HapMap format
model = "BLINK",
PCA.total = 0,
file.output = TRUE,
Major.allele.zero = TRUE
)
Do you suggest converting the genotype data to numeric format separately before running GWAS?

I will also make sure that we use the correct source file.

Thanks again,
Petros

[jiabowang]

Yes, please convert your HapMap file to numeric format at first. Then check whether genotype is correct.
At last, run GAPIT with correct genotype with phenotype. So that the result can be considered as correct.

[PetrosRoth]

Dear Jiabo,

Thank you very much for your help with this.

I followed the steps you suggested, but I noticed another issue after sourcing from https://github.com/jiabowang/GAPIT/blob/master/gapit_functions.txt

The issue I am having now concerns MAF and heterozygosity. As shown in the plot below, I have many SNPs with MAF = 0.5 and heterozygosity = 1. It’s worth noting that when I ran GAPIT in the older version, the graph did not look like this.

Upon checking the tables and genotypic files, I found that this is happening to all monomorphic markers. Below, you can find the table with heterozygosity values and MAF as exported by GAPIT, along with the respective genotypes based on the GD file. Therefore it appears that all monomorphic SNPs got MAF=0.5 and heterozygosity=1.

Could you please look into this?

Thank you in advance for your kind help!

Best regards,
Petros

[jiabowang]

Hi Petros,
Thanks for your figure. That is an estimation mistake for those monomorphic genotype. Now I have fixed it. Please try again with the new GAPIT function.
https://raw.githubusercontent.com/jiabowang/GAPIT/refs/heads/master/gapit_functions.txt